Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/mrsladoje/sweet-searchnpx agentmods add skills/mrsladoje/sweet-search/sweet-indexWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/mrsladoje/sweet-search/sweet-index)<a href="https://agentmods.dev/skills/mrsladoje/sweet-search/sweet-index"><img src="https://agentmods.dev/badge/skills/mrsladoje/sweet-search/sweet-index/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mrsladoje/sweet-search/sweet-index"><img src="https://agentmods.dev/badge/skills/mrsladoje/sweet-search/sweet-index.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00083 | $0.00993 |
| Opus 5 | $0.00042 | $0.00496 |
| Sonnet 5 | $0.00017 | $0.00199 |
| Haiku 4.5 | $0.00008 | $0.00099 |
Grade A, and why
sweet-index scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/sweet-index — Index the Codebase
<default_to_action>
When the user invokes /sweet-index, run the full-profile indexing command
immediately. Do not ask clarifying questions — the indexer is idempotent, safe
to re-run, and handles incremental vs full reindex automatically.
</default_to_action>
What this does
Runs core/indexing/index-codebase-v21.js with the --full flag so every
artifact is rebuilt from scratch. The indexer itself manages the model
lifecycle end-to-end:
- Kill resident ORT CPU models — prevents memory contention and mutex fighting with the GPU models about to be loaded.
- Detect best backend via
hardware-capability.js—coreml-cascadeon M3+ Apple Silicon,candle-metalon M1/M2,candle-cudaon Linux + NVIDIA, and the optimized ORT INT8 CPU path on any host with no usable accelerator (no GPU models are loaded there). - Load GPU models + warmup forward pass (accelerator hosts only) — compiles Metal pipelines, CoreML variant bundles, and BLAS thread pools so the first indexing batch pays no cold-start cost.
- Index the codebase — code graph, vector embeddings, HNSW, late-interaction index, quantized artifacts, sparse-gram index.
- Kill GPU models — releases Metal queues and Neural Engine.
- Load + warmup ORT CPU models — both embedding and LI get one dummy forward pass so the first query after indexing is warm.
On small-changeset incremental runs (under 20 files) — and on any host with no usable accelerator — the indexer skips the GPU swap entirely and indexes on the optimized ORT INT8 CPU path.
Usage
node core/indexing/index-codebase-v21.js --full
Or via npm script:
npm run index:full
What to report
After the command completes, pick out these lines from stderr:
GPU index pool armed (<backend>)→ confirms which backend was usedembed=<load>+<warm>ms, li=<load>+<warm>ms→ prewarm timingsCPU models warmed for queries: load=…ms, warm=…ms (embed=ok, li=ok)→ confirms ORT CPU is armed for subsequent searchesINDEXING COMPLETE (FULL)withDuration,Files indexed,Entities,Relationships→ headline stats
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 95 lines · 83 tokens per session scan A ea56d496cc1e
sweet-index is a skill published in the GitHub repository mrsladoje/sweet-search (21 stars, last pushed yesterday), licensed Apache-2.0. It adds 83 tokens to every session and 993 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
ort
ONNX Runtime in Rust via the ort crate (2.x): loading sessions, configuring CPU/CoreML/CUDA execution providers, tensor I/O with ndarray, async-safe spawnblocking wrapping, global thread-pool init, and debugging provider/opset issues.
vector-search-workflows
Vector search indexing and querying workflows using MCP Vector Search, including setup, reindexing, auto-index strategies, and MCP integration.
arrowspace
Spectral vector search using graph Laplacian eigenstructure. Use when cosine/L2 similarity misses latent structure in your embeddings.
sentence-transformers
Framework for state-of-the-art sentence, text, and image embeddings. Provides 5000+ pre-trained models for semantic similarity, clustering, and retrieval. Supports multilingual, domain-specific, and multimodal models. Use for generating embeddings for RAG, semantic search, or similarity tasks. Best for production…
llm-ops
LLM Operations -- RAG, embeddings, vector databases, fine-tuning, prompt engineering avancado, custos de LLM, evals de qualidade e arquiteturas de IA para producao.
nemotron-retrieval-recipes
Use when planning, debugging, tuning, evaluating, exporting, or deploying public Nemotron embed/rerank retrieval recipes.